kordoc 4.18.7 → 4.18.9

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (123) hide show
  1. package/README.md +74 -561
  2. package/dist/{-3AJPADNH.js → -OOORB3ZM.js} +11 -11
  3. package/dist/{chunk-6FGSSCXN.js → chunk-2ZATWIWX.js} +3 -3
  4. package/dist/{chunk-X6RQA2ZJ.js → chunk-44LAZYCW.js} +2 -2
  5. package/dist/{chunk-MKARDNWJ.js → chunk-5K7M5QCG.js} +2 -2
  6. package/dist/{chunk-32H2UKEZ.js → chunk-5XMZU77R.js} +2 -2
  7. package/dist/{chunk-HR7QNQOQ.cjs → chunk-5Z5VIVZ4.cjs} +3 -3
  8. package/dist/{chunk-HR7QNQOQ.cjs.map → chunk-5Z5VIVZ4.cjs.map} +1 -1
  9. package/dist/{chunk-M6UIEUUU.cjs → chunk-A66VFXSS.cjs} +12 -12
  10. package/dist/{chunk-M6UIEUUU.cjs.map → chunk-A66VFXSS.cjs.map} +1 -1
  11. package/dist/{chunk-T4TXPUOX.js → chunk-BBMS5Z7N.js} +2 -2
  12. package/dist/{chunk-MPUJQOCK.js → chunk-CGRLQJC2.js} +27 -4
  13. package/dist/chunk-CGRLQJC2.js.map +1 -0
  14. package/dist/{chunk-TQN4QRL7.cjs → chunk-FFUNCANU.cjs} +47 -24
  15. package/dist/chunk-FFUNCANU.cjs.map +1 -0
  16. package/dist/{chunk-2M7WLIUZ.js → chunk-H3MGECXS.js} +2 -2
  17. package/dist/{chunk-AOOQK3RX.js → chunk-IPJ5SVNH.js} +2 -2
  18. package/dist/{chunk-UHK47KXY.js → chunk-J6KBWT5D.js} +89 -10
  19. package/dist/chunk-J6KBWT5D.js.map +1 -0
  20. package/dist/{chunk-IM77QPKN.js → chunk-L3YW7ATL.js} +2 -2
  21. package/dist/{chunk-JXVVOSWW.js → chunk-L65DR4EY.js} +2 -2
  22. package/dist/{chunk-LUZF5ESP.js → chunk-LZJ4HHFE.js} +4 -4
  23. package/dist/{chunk-U4MI7XTO.js → chunk-MSGQ6DFT.js} +2 -2
  24. package/dist/{chunk-EB5T5SC6.js → chunk-MW2G24XL.js} +2 -2
  25. package/dist/{chunk-LFB3XTIK.js → chunk-NEQC7PV5.js} +2 -2
  26. package/dist/{chunk-AKQRKMY5.js → chunk-NMKCOSXN.js} +2 -2
  27. package/dist/{chunk-NCVMV4MR.js → chunk-O345LTOL.js} +3 -3
  28. package/dist/{chunk-5DKDP42Q.js → chunk-OJGHM3YD.js} +2 -2
  29. package/dist/{chunk-YROJFVJI.js → chunk-S7ED7QP2.js} +28 -5
  30. package/dist/chunk-S7ED7QP2.js.map +1 -0
  31. package/dist/{chunk-SD5DGUD4.js → chunk-SO66Q6ST.js} +2 -2
  32. package/dist/{chunk-ZCOZENWU.cjs → chunk-SOIGY5LM.cjs} +2 -2
  33. package/dist/{chunk-ZCOZENWU.cjs.map → chunk-SOIGY5LM.cjs.map} +1 -1
  34. package/dist/{chunk-5EXCACRC.cjs → chunk-VMGLV3P3.cjs} +3 -3
  35. package/dist/{chunk-5EXCACRC.cjs.map → chunk-VMGLV3P3.cjs.map} +1 -1
  36. package/dist/{chunk-7ELJ2VQW.js → chunk-XB6VRTM3.js} +2 -2
  37. package/dist/{chunk-T2VQWNMK.cjs → chunk-XMULZUPH.cjs} +6 -6
  38. package/dist/{chunk-T2VQWNMK.cjs.map → chunk-XMULZUPH.cjs.map} +1 -1
  39. package/dist/{chunk-KLOQU3SZ.js → chunk-ZMHMIXTU.js} +6 -6
  40. package/dist/chunks-2AMQTIMX.js +10 -0
  41. package/dist/cli.js +26 -26
  42. package/dist/{formula-WOW5NIIX.cjs → formula-5EGP7AVC.cjs} +9 -9
  43. package/dist/{formula-WOW5NIIX.cjs.map → formula-5EGP7AVC.cjs.map} +1 -1
  44. package/dist/{formula-BDPM3Q37.js → formula-CLVKOQCO.js} +4 -4
  45. package/dist/{formula-K2CIXI4G.js → formula-KOOZT2QS.js} +3 -3
  46. package/dist/{image-ocr-IGO6GXW4.js → image-ocr-AKSRUDIT.js} +9 -9
  47. package/dist/{image-ocr-PXFOI5HL.cjs → image-ocr-BHJFBQ2T.cjs} +15 -15
  48. package/dist/{image-ocr-PXFOI5HL.cjs.map → image-ocr-BHJFBQ2T.cjs.map} +1 -1
  49. package/dist/{image-ocr-SACGXUVC.js → image-ocr-LLMYXDAL.js} +6 -6
  50. package/dist/index.cjs +460 -381
  51. package/dist/index.cjs.map +1 -1
  52. package/dist/index.js +84 -5
  53. package/dist/index.js.map +1 -1
  54. package/dist/mcp.js +22 -22
  55. package/dist/{model-bundle-Y64X36ZK.js → model-bundle-NT6KCK6Y.js} +5 -5
  56. package/dist/{models-NMOGNW7R.js → models-SWD6AIAB.js} +5 -5
  57. package/dist/{parser-CH4MBHEZ.js → parser-ANGCUOPB.js} +10 -9
  58. package/dist/{parser-CZWBWP7H.js.map → parser-ANGCUOPB.js.map} +1 -1
  59. package/dist/{parser-NMWMCBWM.cjs → parser-N5MYD4XA.cjs} +127 -126
  60. package/dist/parser-N5MYD4XA.cjs.map +1 -0
  61. package/dist/{parser-CZWBWP7H.js → parser-RYDOM6KM.js} +13 -12
  62. package/dist/{parser-CH4MBHEZ.js.map → parser-RYDOM6KM.js.map} +1 -1
  63. package/dist/pdf-ocr-B23SDZDD.js +20 -0
  64. package/dist/pdf-ocr-RX3CQENG.cjs +17 -0
  65. package/dist/{pdf-ocr-RPBTK6XS.cjs.map → pdf-ocr-RX3CQENG.cjs.map} +1 -1
  66. package/dist/pdf-ocr-SMFPZJGJ.js +16 -0
  67. package/dist/{profile-io-UFU7APWM.js → profile-io-CZE4YTYD.js} +3 -3
  68. package/dist/{rasterize-NL6RCQWN.js → rasterize-KBBEPWJN.js} +3 -3
  69. package/dist/{redact-doc-FEM3AGUQ.js → redact-doc-K7HCORXE.js} +11 -11
  70. package/dist/{render-WDCKVMGY.js → render-TXGEQCW6.js} +7 -7
  71. package/dist/seal-QBDRMEQ3.js +10 -0
  72. package/dist/{setup-E3E4LDKQ.js → setup-EHCY4PGN.js} +3 -3
  73. package/dist/visual-66VQETYB.js +20 -0
  74. package/dist/{watch-6UYWJ6UG.js → watch-54O2O32S.js} +12 -12
  75. package/package.json +1 -1
  76. package/dist/chunk-MPUJQOCK.js.map +0 -1
  77. package/dist/chunk-TQN4QRL7.cjs.map +0 -1
  78. package/dist/chunk-UHK47KXY.js.map +0 -1
  79. package/dist/chunk-YROJFVJI.js.map +0 -1
  80. package/dist/chunks-JOLLBBQL.js +0 -10
  81. package/dist/parser-NMWMCBWM.cjs.map +0 -1
  82. package/dist/pdf-ocr-BLU34LTG.js +0 -16
  83. package/dist/pdf-ocr-L4LJUSZ3.js +0 -20
  84. package/dist/pdf-ocr-RPBTK6XS.cjs +0 -17
  85. package/dist/seal-XHUWDRIW.js +0 -10
  86. package/dist/visual-74WYDXEK.js +0 -20
  87. /package/dist/{-3AJPADNH.js.map → -OOORB3ZM.js.map} +0 -0
  88. /package/dist/{chunk-6FGSSCXN.js.map → chunk-2ZATWIWX.js.map} +0 -0
  89. /package/dist/{chunk-X6RQA2ZJ.js.map → chunk-44LAZYCW.js.map} +0 -0
  90. /package/dist/{chunk-MKARDNWJ.js.map → chunk-5K7M5QCG.js.map} +0 -0
  91. /package/dist/{chunk-32H2UKEZ.js.map → chunk-5XMZU77R.js.map} +0 -0
  92. /package/dist/{chunk-T4TXPUOX.js.map → chunk-BBMS5Z7N.js.map} +0 -0
  93. /package/dist/{chunk-2M7WLIUZ.js.map → chunk-H3MGECXS.js.map} +0 -0
  94. /package/dist/{chunk-AOOQK3RX.js.map → chunk-IPJ5SVNH.js.map} +0 -0
  95. /package/dist/{chunk-IM77QPKN.js.map → chunk-L3YW7ATL.js.map} +0 -0
  96. /package/dist/{chunk-JXVVOSWW.js.map → chunk-L65DR4EY.js.map} +0 -0
  97. /package/dist/{chunk-LUZF5ESP.js.map → chunk-LZJ4HHFE.js.map} +0 -0
  98. /package/dist/{chunk-U4MI7XTO.js.map → chunk-MSGQ6DFT.js.map} +0 -0
  99. /package/dist/{chunk-EB5T5SC6.js.map → chunk-MW2G24XL.js.map} +0 -0
  100. /package/dist/{chunk-LFB3XTIK.js.map → chunk-NEQC7PV5.js.map} +0 -0
  101. /package/dist/{chunk-AKQRKMY5.js.map → chunk-NMKCOSXN.js.map} +0 -0
  102. /package/dist/{chunk-NCVMV4MR.js.map → chunk-O345LTOL.js.map} +0 -0
  103. /package/dist/{chunk-5DKDP42Q.js.map → chunk-OJGHM3YD.js.map} +0 -0
  104. /package/dist/{chunk-SD5DGUD4.js.map → chunk-SO66Q6ST.js.map} +0 -0
  105. /package/dist/{chunk-7ELJ2VQW.js.map → chunk-XB6VRTM3.js.map} +0 -0
  106. /package/dist/{chunk-KLOQU3SZ.js.map → chunk-ZMHMIXTU.js.map} +0 -0
  107. /package/dist/{chunks-JOLLBBQL.js.map → chunks-2AMQTIMX.js.map} +0 -0
  108. /package/dist/{formula-BDPM3Q37.js.map → formula-CLVKOQCO.js.map} +0 -0
  109. /package/dist/{formula-K2CIXI4G.js.map → formula-KOOZT2QS.js.map} +0 -0
  110. /package/dist/{image-ocr-IGO6GXW4.js.map → image-ocr-AKSRUDIT.js.map} +0 -0
  111. /package/dist/{image-ocr-SACGXUVC.js.map → image-ocr-LLMYXDAL.js.map} +0 -0
  112. /package/dist/{model-bundle-Y64X36ZK.js.map → model-bundle-NT6KCK6Y.js.map} +0 -0
  113. /package/dist/{models-NMOGNW7R.js.map → models-SWD6AIAB.js.map} +0 -0
  114. /package/dist/{pdf-ocr-BLU34LTG.js.map → pdf-ocr-B23SDZDD.js.map} +0 -0
  115. /package/dist/{pdf-ocr-L4LJUSZ3.js.map → pdf-ocr-SMFPZJGJ.js.map} +0 -0
  116. /package/dist/{profile-io-UFU7APWM.js.map → profile-io-CZE4YTYD.js.map} +0 -0
  117. /package/dist/{rasterize-NL6RCQWN.js.map → rasterize-KBBEPWJN.js.map} +0 -0
  118. /package/dist/{redact-doc-FEM3AGUQ.js.map → redact-doc-K7HCORXE.js.map} +0 -0
  119. /package/dist/{render-WDCKVMGY.js.map → render-TXGEQCW6.js.map} +0 -0
  120. /package/dist/{seal-XHUWDRIW.js.map → seal-QBDRMEQ3.js.map} +0 -0
  121. /package/dist/{setup-E3E4LDKQ.js.map → setup-EHCY4PGN.js.map} +0 -0
  122. /package/dist/{visual-74WYDXEK.js.map → visual-66VQETYB.js.map} +0 -0
  123. /package/dist/{watch-6UYWJ6UG.js.map → watch-54O2O32S.js.map} +0 -0
@@ -4,15 +4,15 @@
4
4
 
5
5
 
6
6
 
7
- var _chunkT2VQWNMKcjs = require('./chunk-T2VQWNMK.cjs');
7
+ var _chunkXMULZUPHcjs = require('./chunk-XMULZUPH.cjs');
8
8
 
9
9
 
10
10
 
11
- var _chunkHR7QNQOQcjs = require('./chunk-HR7QNQOQ.cjs');
11
+ var _chunk5Z5VIVZ4cjs = require('./chunk-5Z5VIVZ4.cjs');
12
12
 
13
13
 
14
14
 
15
- var _chunkZCOZENWUcjs = require('./chunk-ZCOZENWU.cjs');
15
+ var _chunkSOIGY5LMcjs = require('./chunk-SOIGY5LM.cjs');
16
16
 
17
17
  // src/ocr/models.ts
18
18
  var _path = require('path');
@@ -40,10 +40,10 @@ var OCR_REC_DICT = {
40
40
  };
41
41
  var ALL_OCR_MODELS = [OCR_DET_MODEL, OCR_REC_MODEL, OCR_REC_DICT];
42
42
  function getOcrModelsDir() {
43
- return _chunkHR7QNQOQcjs.getModelsDir.call(void 0, "ppocr");
43
+ return _chunk5Z5VIVZ4cjs.getModelsDir.call(void 0, "ppocr");
44
44
  }
45
45
  async function ensureOcrModels(onProgress) {
46
- return _chunkHR7QNQOQcjs.ensureModelsIn.call(void 0, getOcrModelsDir(), ALL_OCR_MODELS, onProgress);
46
+ return _chunk5Z5VIVZ4cjs.ensureModelsIn.call(void 0, getOcrModelsDir(), ALL_OCR_MODELS, onProgress);
47
47
  }
48
48
  async function ocrModelsCached() {
49
49
  const dir = getOcrModelsDir();
@@ -2091,7 +2091,7 @@ function tagScripts(text, lines) {
2091
2091
  const [s, e, kind] = marks[m];
2092
2092
  out = out.slice(0, s) + `<${kind}>` + out.slice(s, e) + `</${kind}>` + out.slice(e);
2093
2093
  }
2094
- return _chunkT2VQWNMKcjs.tidyScriptTags.call(void 0, out);
2094
+ return _chunkXMULZUPHcjs.tidyScriptTags.call(void 0, out);
2095
2095
  }
2096
2096
 
2097
2097
  // src/pdf/line-wrap.ts
@@ -2242,6 +2242,7 @@ function wrapJoiner(prevText, nextText, lex) {
2242
2242
  return " ";
2243
2243
  }
2244
2244
  var PARA_LAST_LINE = /* @__PURE__ */ new WeakMap();
2245
+ var PAGE_BREAK_JOINS = /* @__PURE__ */ new WeakMap();
2245
2246
  function joinPageBreakWraps(blocks, lex) {
2246
2247
  const pageRight = /* @__PURE__ */ new Map();
2247
2248
  for (const b of blocks) {
@@ -2256,13 +2257,34 @@ function joinPageBreakWraps(blocks, lex) {
2256
2257
  if (!last || fs <= 0 || (_nullishCoalesce(pageRight.get(a.pageNumber), () => ( Infinity))) - last.right >= BODY_FULL_TOL * fs) continue;
2257
2258
  if (last.width < BODY_MIN_WIDTH_EM * fs || Math.abs((_nullishCoalesce(_optionalChain([b, 'access', _56 => _56.style, 'optionalAccess', _57 => _57.fontSize]), () => ( 0))) - fs) > 0.15 * fs) continue;
2258
2259
  if (startsNewItem(a.text, b.text)) continue;
2259
- a.text += wrapJoiner(a.text, b.text, lex) + b.text;
2260
+ const head = a.text, joiner = wrapJoiner(a.text, b.text, lex);
2261
+ a.text += joiner + b.text;
2262
+ const later = b.bbox && PAGE_BREAK_JOINS.get(b.bbox) || [];
2263
+ PAGE_BREAK_JOINS.set(a.bbox, [{ head, gap: joiner.length, tail: b }, ...later.map((c) => ({ ...c, head: head + joiner + c.head }))]);
2260
2264
  const bl = b.bbox && PARA_LAST_LINE.get(b.bbox);
2261
2265
  if (bl) PARA_LAST_LINE.set(a.bbox, bl);
2262
2266
  else PARA_LAST_LINE.delete(a.bbox);
2263
2267
  blocks.splice(i, 1);
2264
2268
  }
2265
2269
  }
2270
+ function splitPageBreakWraps(blocks) {
2271
+ return blocks.flatMap((b) => {
2272
+ const cuts = b.bbox && PAGE_BREAK_JOINS.get(b.bbox);
2273
+ const text = b.text;
2274
+ if (!cuts || !text) return [b];
2275
+ const out = [];
2276
+ let from = 0, cur = b;
2277
+ for (const c of cuts) {
2278
+ if (c.head.length < from || !text.startsWith(c.head)) break;
2279
+ out.push({ ...cur, text: text.slice(from, c.head.length) });
2280
+ from = c.head.length + c.gap;
2281
+ cur = c.tail;
2282
+ }
2283
+ if (!out.length) return [b];
2284
+ out.push({ ...cur, text: text.slice(from) });
2285
+ return out;
2286
+ });
2287
+ }
2266
2288
  var HANGING = /[,。、:;!?)」』】〕》〉.]\s*$/;
2267
2289
  var MULTI_LEVEL_NUMBER = /^\d{1,2}(?:\.\d{1,3})+\.?\s/;
2268
2290
  var BODY_FULL_TOL = 0.25;
@@ -3644,9 +3666,9 @@ function detectHeadings(blocks, medianFontSize) {
3644
3666
  if (/^\d+$/.test(text) && !isChapterNumber(block, blocks[bi + 1], medianFontSize)) continue;
3645
3667
  const ratio = block.style.fontSize / medianFontSize;
3646
3668
  let level = 0;
3647
- if (ratio >= _chunkT2VQWNMKcjs.HEADING_RATIO_H1) level = 1;
3648
- else if (ratio >= _chunkT2VQWNMKcjs.HEADING_RATIO_H2) level = 2;
3649
- else if (ratio >= _chunkT2VQWNMKcjs.HEADING_RATIO_H3) level = 3;
3669
+ if (ratio >= _chunkXMULZUPHcjs.HEADING_RATIO_H1) level = 1;
3670
+ else if (ratio >= _chunkXMULZUPHcjs.HEADING_RATIO_H2) level = 2;
3671
+ else if (ratio >= _chunkXMULZUPHcjs.HEADING_RATIO_H3) level = 3;
3650
3672
  if (level > 0) {
3651
3673
  block.type = "heading";
3652
3674
  block.level = level;
@@ -3659,7 +3681,7 @@ function isChapterNumber(block, next, medianFontSize) {
3659
3681
  const nextSize = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _96 => _96.style, 'optionalAccess', _97 => _97.fontSize]), () => ( 0));
3660
3682
  const text = _nullishCoalesce(_optionalChain([next, 'optionalAccess', _98 => _98.text, 'optionalAccess', _99 => _99.trim, 'call', _100 => _100()]), () => ( ""));
3661
3683
  const a = block.bbox, b = _optionalChain([next, 'optionalAccess', _101 => _101.bbox]);
3662
- return /^\d{1,2}$/.test(_nullishCoalesce(_optionalChain([block, 'access', _102 => _102.text, 'optionalAccess', _103 => _103.trim, 'call', _104 => _104()]), () => ( ""))) && size >= medianFontSize * 2.5 && !!next && (next.type === "paragraph" || next.type === "heading") && next.pageNumber === block.pageNumber && nextSize >= medianFontSize * _chunkT2VQWNMKcjs.HEADING_RATIO_H1 && text.length > 0 && text.length <= 80 && !/^[\d\s.]+$/.test(text) && !!a && !!b && a.y > b.y && a.y - (b.y + b.height) <= size * 2;
3684
+ return /^\d{1,2}$/.test(_nullishCoalesce(_optionalChain([block, 'access', _102 => _102.text, 'optionalAccess', _103 => _103.trim, 'call', _104 => _104()]), () => ( ""))) && size >= medianFontSize * 2.5 && !!next && (next.type === "paragraph" || next.type === "heading") && next.pageNumber === block.pageNumber && nextSize >= medianFontSize * _chunkXMULZUPHcjs.HEADING_RATIO_H1 && text.length > 0 && text.length <= 80 && !/^[\d\s.]+$/.test(text) && !!a && !!b && a.y > b.y && a.y - (b.y + b.height) <= size * 2;
3663
3685
  }
3664
3686
  function mergeStackedHeadingLines(blocks, medianFontSize) {
3665
3687
  for (let i = 0; i < blocks.length - 1; ) {
@@ -5971,16 +5993,16 @@ function isProseSpread(items) {
5971
5993
  for (let i = 1; i < sorted.length; i++) {
5972
5994
  gaps.push(sorted[i].x - (sorted[i - 1].x + sorted[i - 1].w));
5973
5995
  }
5974
- const maxGap = _chunkZCOZENWUcjs.safeMax.call(void 0, gaps);
5996
+ const maxGap = _chunkSOIGY5LMcjs.safeMax.call(void 0, gaps);
5975
5997
  const avgLen = items.reduce((s, i) => s + i.text.length, 0) / items.length;
5976
5998
  if (items.length >= 4 && maxGap < 40 && avgLen < 5) return true;
5977
- const minGap = _chunkZCOZENWUcjs.safeMin.call(void 0, gaps);
5999
+ const minGap = _chunkSOIGY5LMcjs.safeMin.call(void 0, gaps);
5978
6000
  return minGap > 0 && maxGap - minGap <= Math.max(2, maxGap * 0.15) && items.filter((i) => /^\P{L}*\p{Ll}/u.test(i.text)).length * 2 >= items.length;
5979
6001
  }
5980
6002
  function detectColumns(yLines) {
5981
6003
  const allItems = yLines.flat();
5982
6004
  if (allItems.length === 0) return null;
5983
- const pageWidth = _chunkZCOZENWUcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkZCOZENWUcjs.safeMin.call(void 0, allItems.map((i) => i.x));
6005
+ const pageWidth = _chunkSOIGY5LMcjs.safeMax.call(void 0, allItems.map((i) => i.x + i.w)) - _chunkSOIGY5LMcjs.safeMin.call(void 0, allItems.map((i) => i.x));
5984
6006
  if (pageWidth < 100) return null;
5985
6007
  let bigoLineIdx = -1;
5986
6008
  for (let i = 0; i < yLines.length; i++) {
@@ -6450,12 +6472,12 @@ function recordClipCellEdges(grids, horizontals, verticals, nonRules) {
6450
6472
  }
6451
6473
  function takeClipCellEdges(from, to) {
6452
6474
  const e = CLIP_CELL_EDGES.get(from);
6453
- if (e) _chunkT2VQWNMKcjs.CELL_EDGES.set(to, e);
6475
+ if (e) _chunkXMULZUPHcjs.CELL_EDGES.set(to, e);
6454
6476
  }
6455
6477
  function joinCellEdges(a, b) {
6456
- const u = _chunkT2VQWNMKcjs.CELL_EDGES.get(a), d = _chunkT2VQWNMKcjs.CELL_EDGES.get(b);
6478
+ const u = _chunkXMULZUPHcjs.CELL_EDGES.get(a), d = _chunkXMULZUPHcjs.CELL_EDGES.get(b);
6457
6479
  if (!u || !d) return;
6458
- _chunkT2VQWNMKcjs.CELL_EDGES.set(a, { t: u.t, b: d.b, l: u.l || d.l, r: u.r || d.r });
6480
+ _chunkXMULZUPHcjs.CELL_EDGES.set(a, { t: u.t, b: d.b, l: u.l || d.l, r: u.r || d.r });
6459
6481
  }
6460
6482
 
6461
6483
  // src/pdf/table-unit-row.ts
@@ -6471,7 +6493,7 @@ function rebuildUnitLine(items, grid) {
6471
6493
  function prependUnitRow(cells, cols, line, usedItems) {
6472
6494
  const text = cleanCellText(line.map((item) => item.text).join(""));
6473
6495
  cells.unshift(Array.from({ length: cols }, (_, c) => ({ text: c === 0 ? text : "", colSpan: c === 0 ? cols : 1, rowSpan: 1 })));
6474
- for (const cell of cells[0]) _chunkT2VQWNMKcjs.CELL_EDGES.set(cell, NO_EDGES);
6496
+ for (const cell of cells[0]) _chunkXMULZUPHcjs.CELL_EDGES.set(cell, NO_EDGES);
6475
6497
  for (const item of line) usedItems.add(item);
6476
6498
  }
6477
6499
  function attachUnitRow(items, grid, cells, cols, usedItems) {
@@ -7001,7 +7023,7 @@ function recordFrameTitle(table, source, pageNum) {
7001
7023
  const [label, spacer, title] = cells;
7002
7024
  if (!/^(?:붙임|별첨|첨부|부록)\s*\d{1,3}$/.test(label.text.trim()) || spacer.text.trim() || !title.text.trim()) return;
7003
7025
  if (cells.some((c) => {
7004
- const e = _chunkT2VQWNMKcjs.CELL_EDGES.get(c);
7026
+ const e = _chunkXMULZUPHcjs.CELL_EDGES.get(c);
7005
7027
  return !e || e.t || e.b || e.l || e.r;
7006
7028
  })) return;
7007
7029
  const xs = TABLE_COLXS.get(table), style = dominantStyle(source), fs = _nullishCoalesce(_optionalChain([style, 'optionalAccess', _324 => _324.fontSize]), () => ( 0));
@@ -7913,7 +7935,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
7913
7935
  groupSizes.push(1);
7914
7936
  }
7915
7937
  const allY = remaining.map((i) => i.y);
7916
- const pageH = _chunkZCOZENWUcjs.safeMax.call(void 0, allY) - _chunkZCOZENWUcjs.safeMin.call(void 0, allY);
7938
+ const pageH = _chunkSOIGY5LMcjs.safeMax.call(void 0, allY) - _chunkSOIGY5LMcjs.safeMin.call(void 0, allY);
7917
7939
  const gapThreshold = Math.max(15, pageH * 0.03);
7918
7940
  const sides = frameColumnTextBands(remaining, spanning, gx);
7919
7941
  const textBlocks = [];
@@ -7929,7 +7951,7 @@ function extractBlocksWithGrids(items, pageNum, pageWidth, pageHeight, grids, ho
7929
7951
  finalTextBlocks.push(...detectListBlocks(textBlocks));
7930
7952
  } else {
7931
7953
  const allY = remaining.map((i) => i.y);
7932
- const pageH = _chunkZCOZENWUcjs.safeMax.call(void 0, allY) - _chunkZCOZENWUcjs.safeMin.call(void 0, allY);
7954
+ const pageH = _chunkSOIGY5LMcjs.safeMax.call(void 0, allY) - _chunkSOIGY5LMcjs.safeMin.call(void 0, allY);
7933
7955
  const groups = xyCutOrder(remaining, Math.max(15, pageH * 0.03));
7934
7956
  const textBlocks = [];
7935
7957
  for (const group of groups) {
@@ -8254,7 +8276,7 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false, detectTable
8254
8276
  blocks.push(...columnTextToBlocks(tableText, pageNum, bbox, dominantStyle(items)));
8255
8277
  } else {
8256
8278
  const allY = items.map((i) => i.y);
8257
- const pageHeight = _chunkZCOZENWUcjs.safeMax.call(void 0, allY) - _chunkZCOZENWUcjs.safeMin.call(void 0, allY);
8279
+ const pageHeight = _chunkSOIGY5LMcjs.safeMax.call(void 0, allY) - _chunkSOIGY5LMcjs.safeMin.call(void 0, allY);
8258
8280
  const gapThreshold = Math.max(15, pageHeight * 0.03);
8259
8281
  const orderedGroups = proseCutX !== null ? splitTwoColumnProse(items, proseCutX) : xyCutOrder(items, gapThreshold);
8260
8282
  for (const group of orderedGroups) {
@@ -8331,5 +8353,6 @@ function extractPageBlocksFallback(items, pageNum, fullPage = false, detectTable
8331
8353
 
8332
8354
 
8333
8355
 
8334
- exports.OCR_DET_MODEL = OCR_DET_MODEL; exports.OCR_REC_MODEL = OCR_REC_MODEL; exports.OCR_REC_DICT = OCR_REC_DICT; exports.getOcrModelsDir = getOcrModelsDir; exports.ensureOcrModels = ensureOcrModels; exports.ocrModelsCached = ocrModelsCached; exports.parseCharacterDict = parseCharacterDict; exports.extractLines = extractLines; exports.extractImageRegions = extractImageRegions; exports.filterHiddenText = filterHiddenText; exports.collapseEvenSpacing = collapseEvenSpacing; exports.computeBBox = computeBBox; exports.dominantStyle = dominantStyle; exports.normalizeItems = normalizeItems; exports.groupByY = groupByY; exports.mergeLineSimple = mergeLineSimple; exports.WrapLexicon = WrapLexicon; exports.startsNewItem = startsNewItem; exports.wrapJoiner = wrapJoiner; exports.joinPageBreakWraps = joinPageBreakWraps; exports.latinSoftWrap = latinSoftWrap; exports.TOC_BLOCKS = TOC_BLOCKS; exports.computeMedianFontSizeFromFreq = computeMedianFontSizeFromFreq; exports.detectHeadings = detectHeadings; exports.mergeStackedHeadingLines = mergeStackedHeadingLines; exports.detectTypographyHeadings = detectTypographyHeadings; exports.detectDocumentStyleHeadings = detectDocumentStyleHeadings; exports.detectSiblingStyleHeadings = detectSiblingStyleHeadings; exports.detectRepeatedPageLabels = detectRepeatedPageLabels; exports.detectPageLeadHeadings = detectPageLeadHeadings; exports.refineDocumentStyleHeadings = refineDocumentStyleHeadings; exports.detectMarkerHeadings = detectMarkerHeadings; exports.detectTableCaptions = detectTableCaptions; exports.detectKoreanListBlocks = detectKoreanListBlocks; exports.removeHeaderFooterBlocks = removeHeaderFooterBlocks; exports.CLIP_TABLES = CLIP_TABLES; exports.FRAME_TITLE_BLOCKS = FRAME_TITLE_BLOCKS; exports.TABLE_COLXS = TABLE_COLXS; exports.TABLE_ROWYS = TABLE_ROWYS; exports.TABLE_TAIL = TABLE_TAIL; exports.PART_COLXS = PART_COLXS; exports.EMPTY_PARTS = EMPTY_PARTS; exports.IMAGE_CELLS = IMAGE_CELLS; exports.FILLER_CELLS = FILLER_CELLS; exports.CONT_PARTS = CONT_PARTS; exports.CELL_LINES = CELL_LINES; exports.ROW_RULES = ROW_RULES; exports.trimTrailingEmptyTableCols = trimTrailingEmptyTableCols; exports.NO_EDGES = NO_EDGES; exports.joinCellEdges = joinCellEdges; exports.FRAME_READING_UNITS = FRAME_READING_UNITS; exports.recordFrameReadingUnit = recordFrameReadingUnit; exports.restoreImageBullets = restoreImageBullets; exports.detectRightArrowRegions = detectRightArrowRegions; exports.removeSideTabs = removeSideTabs; exports.extractPageBlocksWithLines = extractPageBlocksWithLines;
8335
- //# sourceMappingURL=chunk-TQN4QRL7.cjs.map
8356
+
8357
+ exports.OCR_DET_MODEL = OCR_DET_MODEL; exports.OCR_REC_MODEL = OCR_REC_MODEL; exports.OCR_REC_DICT = OCR_REC_DICT; exports.getOcrModelsDir = getOcrModelsDir; exports.ensureOcrModels = ensureOcrModels; exports.ocrModelsCached = ocrModelsCached; exports.parseCharacterDict = parseCharacterDict; exports.extractLines = extractLines; exports.extractImageRegions = extractImageRegions; exports.filterHiddenText = filterHiddenText; exports.collapseEvenSpacing = collapseEvenSpacing; exports.computeBBox = computeBBox; exports.dominantStyle = dominantStyle; exports.normalizeItems = normalizeItems; exports.groupByY = groupByY; exports.mergeLineSimple = mergeLineSimple; exports.WrapLexicon = WrapLexicon; exports.startsNewItem = startsNewItem; exports.wrapJoiner = wrapJoiner; exports.joinPageBreakWraps = joinPageBreakWraps; exports.splitPageBreakWraps = splitPageBreakWraps; exports.latinSoftWrap = latinSoftWrap; exports.TOC_BLOCKS = TOC_BLOCKS; exports.computeMedianFontSizeFromFreq = computeMedianFontSizeFromFreq; exports.detectHeadings = detectHeadings; exports.mergeStackedHeadingLines = mergeStackedHeadingLines; exports.detectTypographyHeadings = detectTypographyHeadings; exports.detectDocumentStyleHeadings = detectDocumentStyleHeadings; exports.detectSiblingStyleHeadings = detectSiblingStyleHeadings; exports.detectRepeatedPageLabels = detectRepeatedPageLabels; exports.detectPageLeadHeadings = detectPageLeadHeadings; exports.refineDocumentStyleHeadings = refineDocumentStyleHeadings; exports.detectMarkerHeadings = detectMarkerHeadings; exports.detectTableCaptions = detectTableCaptions; exports.detectKoreanListBlocks = detectKoreanListBlocks; exports.removeHeaderFooterBlocks = removeHeaderFooterBlocks; exports.CLIP_TABLES = CLIP_TABLES; exports.FRAME_TITLE_BLOCKS = FRAME_TITLE_BLOCKS; exports.TABLE_COLXS = TABLE_COLXS; exports.TABLE_ROWYS = TABLE_ROWYS; exports.TABLE_TAIL = TABLE_TAIL; exports.PART_COLXS = PART_COLXS; exports.EMPTY_PARTS = EMPTY_PARTS; exports.IMAGE_CELLS = IMAGE_CELLS; exports.FILLER_CELLS = FILLER_CELLS; exports.CONT_PARTS = CONT_PARTS; exports.CELL_LINES = CELL_LINES; exports.ROW_RULES = ROW_RULES; exports.trimTrailingEmptyTableCols = trimTrailingEmptyTableCols; exports.NO_EDGES = NO_EDGES; exports.joinCellEdges = joinCellEdges; exports.FRAME_READING_UNITS = FRAME_READING_UNITS; exports.recordFrameReadingUnit = recordFrameReadingUnit; exports.restoreImageBullets = restoreImageBullets; exports.detectRightArrowRegions = detectRightArrowRegions; exports.removeSideTabs = removeSideTabs; exports.extractPageBlocksWithLines = extractPageBlocksWithLines;
8358
+ //# sourceMappingURL=chunk-FFUNCANU.cjs.map